CinEro-NoiceAI-XL
lynodie
lynodie
SDXL Checkpoint
·
Uploaded Jan 17, 2026
·
Used 243 times
3h7SW7owS65TZj3q
fZAaSHnu7HqNGH3E
EfNBffudpJ6eonfC
ZztJ4H99GXM98yb9
N9PzF9DDTJp48qoY
ZAoxjKmtJLZSVj5A
7tNk55zxMgeuSyxA
TeaQF2QzBxpaTgpC
WYpa5fTJW5bd3GvQ
Shdvw6WDJkCX63kd
BsbbkTnYh3GoVgm4
2 showcases were hidden due to browsing mode.
Description

Creator: https://civitai.com/user/homoludens

About this version

R7 smol soft [EXPERIMENTAL RELEASE VERSION] = It was trained in Full Fine Tuning mode with BF16 precision. CLIP Skip 2 used during training, but CLIP Skip 1 seems also working and gives more sharp renders. Trained version was quantized into FP8 for lower grade computers and lower VRAM requirements.

Low-weight mixture of R7j RC2 applied for better stability and style inheritance.

Goal of training: more detailed and complex backgrounds; improvements of eyes, fingers and structural consistency in background (straight lines matching); contrast and colors control in prompts. Quick tests show the improvements of eyes, fingers and textures. R7j had too sharp unnatural textures in some cases. This version has the chance of unnatural textures lower.


Rule for steps: 8 steps kinda works but gives smooth picture like LCM sampler but with any samplers; 16 steps starts to give details using contrast samplers like DPM++ 2S A; 25..30 steps sharp enough for single pass render, but small objects like eyes and fingers might be unstable; not yet tested on 40+ steps, but I don't expect significant quality improvement.

Euler A with CFG up to 6.0 and steps from 14 and greater works as LCM sampler for previous (DMD2 versions). Euler A gives low noise, very smooth pictures. Good for Text2Image with consequent HiRes fix using other samplers.

DPM++ 2S A + Karras with CFG near 2.5 and steps 20+ is good for realism and texture rendering/optimal to me as a Text-to-Image.

Related Posts